Skip to content

Our article2 min read

Microsoft MAI Code 2 and Flash: Prices, Who Gets Them and What the Benchmarks Leave Out

Microsoft launched two coding models, MAI Code 2 and MAI Code 2 Flash, at Ignite on 17 November 2026. Here are the promotional prices, who can use them, and why the benchmark claims are Microsoft's own.

Maya, the All About AI writer

By Maya, AI writer
Checked by a person at Zettaz before publishing

ShareFacebookXLinkedIn
In this article

Key takeaways

  • Both models are in GitHub Copilot; Flash also reaches Free and Student users, while Code 2 needs a paid plan, per iTWire.
  • Promotional prices run until 17 May 2027 and then double, according to InfoWorld and iTWire.
  • Microsoft's own benchmarks put Code 2 slightly below GPT-6.1 Sol and Claude Opus 5.5 on average but far cheaper per task.
  • The results are vendor-reported and not independently replicated.

What launched

At Microsoft Ignite on 17 November 2026 the company introduced MAI Code 2 and MAI Code 2 Flash. InfoWorld describes Code 2 as aimed at enterprise coding such as deep debugging, changes across many files and interface work, and Flash as aimed at high-volume jobs such as explaining code, small edits and routine fixes.

Microsoft says the models are not yet tuned for harder problems such as fundamental science research or very long-running agent work, as quoted by InfoWorld and The New Stack.

Prices and plans

Prices come from GitHub's pricing page as cited by InfoWorld and iTWire. MAI Code 2 costs $0.50 per million input tokens and $2.50 per million output tokens during a promotion that runs to 17 May 2027, then $1.00 and $5.00. Flash costs $0.05 input and $0.25 output, rising to $0.10 and $0.50.

Both models are in GitHub Copilot. iTWire, citing GitHub, says Code 2 is for Copilot Pro and other eligible paid plans, including Business and Enterprise, and Flash also reaches Free and Student users. Admins on Business and Enterprise can control access through model policies, and new models are switched on by default unless changed.

What Microsoft's numbers say

InfoWorld relays Microsoft's tests, run inside GitHub Copilot on four coding benchmarks. MAI Code 2 averaged 63.2 per cent, against 64.1 for GPT-6.1 Sol and 65.3 for Claude Opus 5.5. Flash averaged 61 per cent against 55.1 for GPT-6 Luna and 53.9 for Claude Haiku 5.5.

Microsoft's cost per task was $0.309 for Code 2, $0.761 for Sol and $6.306 for Opus; Flash was $0.052. Microsoft's own cost figures use the 50 per cent promotional price, so the gap narrows when the promotion ends.

Where sources differ

The New Stack says Flash matches Haiku and Luna, while InfoWorld's table shows Flash ahead on average. The New Stack also puts Flash at 20 per cent of Haiku's cost, whereas InfoWorld's per-task figures work out closer to 16 per cent. We could not open Microsoft's announcement to settle either point.

What to weigh before switching

  • Every benchmark and cost figure is Microsoft's; iTWire says they are not independently replicated.
  • Analyst Sanchit Vir Gogia, quoted by InfoWorld, advises testing on your own code and tracking cost per merged change and security findings.
  • Pareekh Jain says model cost may be a small part of total development spend, so a large model discount may save far less overall; his example is that a 60 per cent cut in model cost gives only about 6 per cent overall savings if models are 10 per cent of development cost.
  • Model cards rate both models highly on offensive cyber capability, InfoWorld notes.

Still to come

Flash is in limited preview on Microsoft Foundry. Support on other platforms is promised as coming soon, and iTWire says it is not yet established as live. Microsoft also named a security model, MAI Cyber 2 Flash, for its own security products from mid-December. We have not tested any of these models.

Go deeper

Sources

Written by an AI from the sources above and checked by a person before publishing. We have not tested products unless we say so. See our editorial policy or tell us about an error through contact.

ShareFacebookXLinkedIn

Keep reading